cv

Computer Vision Engineer

Dexterity Inc.

• Robust Camera-Robot Hand-eye Calibration in presence of forward-kinematics and measurement noise.
• Characterize and correct depth measurement error in RGB-D cameras.
• Accurate depth prediction/filling using stereo cameras.

Research Engineer Intern

Meta Inc.

• Designed algorithm for multi-session localization using sparse point cloud alignment.
• Developed a pipeline to detect and correct localization errors during multi-session operations of the VR headset.

Research Assistant

Autonomous Field Robotics Lab

• Improved visual SLAM by tightly-coupled fusion of magnetometer measurements using IMU preintegration.
• Designed a switching estimator framework to keep track of a robot’s pose when visual input to VIO is degraded.
• Real-time dense reconstruction in challenging underwater environments.
• Developed algorithm for global accurate 3D mapping of large structures using Visual-Inertial data from GoPro
• Designed real-time deep learning-based 6D pose estimation framework for AUV using only synthetic images.
• Evaluated the performance of different visual-inertial state estimation algorithms in the underwater domain.

PhD

University of South Carolina

Robot Vision, Robust SLAM and Deep Learning

Enhancing Visual Inertial SLAM with Magnetic Measurements

ICRA (2024)

This paper presents an extension to visual inertial odometry (VIO) by introducing tightly-coupled fusion of magnetometer measurements resulting in significant reductions in the orientation error and also in recovery of the true yaw orientation with respect to the magnetic north.

Real-Time Dense 3D Mapping of Underwater Environments

ICRA (2023)

We present a real-time dense 3D reconstruction pipeline for underwater environments.

Deep-URL: Deep Pose Estimation Framework for Underwater Relative Localization

IROS (2020)

We propose a real-time deep learning approach for determining the 6D relative pose of Autonomous Underwater Vehicles (AUV) from a single image.